Underlying Emotions in Political Messages: An Analysis of Their Influence on Interactions and Sharing/Commenting Actions on Facebook

Abstract

The 2022 presidential elections in Colombia resulted in the victory of the first left-wing government in the country’s history, amidst a campaign marked by economic crisis, polarization, widespread circulation of misinformation, and hate speech fueling a populist narrative with a significant emotional component. In this context and considering the centrality of social media in public discourse, this study employed automated natural language processing techniques and statistical analysis to explore the configuration of affective audiences in the electoral contest, examining the specific emotions and implicit connotations within the breadth of campaign messages on the social network Facebook (N = 3,002) and their relationship with user reactions. One contribution of this study was the analysis of reactions as a mediator between implicit emotional valences and interactions (commenting and sharing). Although the implicit negative valence in campaign messages did not show a direct association with interactions, it did show an indirect and significant association through an increase in the number of negative reactions (“anger”), confirming the heuristic role of reactions and suggesting that the affective alignment between valences and reactions is a factor that enhances the impact of posts.